aws-samples/aidlc-discovery
AI-DLC Discovery: run Discovery in five phases, from customer feedback to a Discovery brief, with one skill per job.
Runs AI-DLC Discovery end to end in five phases (Signals, Working Backwards, Prototype, Validate, Handoff), from customer feedback to a Discovery brief the team builds from. Use when the user wants to start or continue Discovery, decide what to build from customer feedback, or asks where they are in Discovery. Calls one Discovery skill per job and stops at each checkpoint for the team.
Go-to-market section for AI-DLC Discovery: drafts first adopters, positioning, business model, and launch plan in Working Backwards, then completes the viability check and the adoption and success table in Validate. Use when the user asks who adopts first, how to position the idea, how adoption will be measured, or wants the GTM or adoption plan.
Handoff phase of AI-DLC Discovery: checks every required artifact exists, then writes the Discovery brief, the single document the team builds from, into the team's repository, with the personas for AI-DLC and the command that starts AI-DLC Inception from the brief. Use when the user wants the Discovery brief, to hand off to engineering, AAP, or AI-DLC, or to finish Discovery.
Synthetic personas for AI-DLC Discovery: generates personas from customer feedback or Signal Briefs, runs the persona critique of a prototype at the end of Prototype, and lets the team talk to a persona or run a persona panel. Use when the user wants to create personas, hear how a persona would react to an idea or prototype, run a persona critique, or hold a panel discussion.
Prototype phase of AI-DLC Discovery: builds a polished, clickable prototype of the core user flow, with several screens shown on a phone and in a desktop browser, in one self-contained HTML file, aimed at the riskiest assumption, then gets the persona critique and iterates with the team. Use when the user wants to build, change, or share a prototype, mockup, or click-dummy for a Discovery use case.
Writes Signal Briefs for AI-DLC Discovery: one-page, anonymized summaries of one customer problem each, created by the team from its own data inside its own environment, so raw data never leaves it. Use before Discovery when the team cannot bring raw customer data, or when asked to write, check, or anonymize a Signal Brief.
Signals phase of AI-DLC Discovery: turns customer feedback, Signal Briefs, or a synthetic dataset into a Signal Analysis Report and testable hypotheses, after the team writes down what it expects to find. Use when the user wants to analyze customer feedback or VoC data, combine Signal Briefs, find the top customer problems with evidence, or start the Signals phase of Discovery.
Follow-up surveys for AI-DLC Discovery: turns the open hypotheses from Validate into a short survey for real users, as an HTML form and as questions ready to paste into Microsoft Forms or Google Forms, then builds a results dashboard from the responses and updates each hypothesis's verdict. Use after Discovery to test open hypotheses or the prototype with real users.
Saves AI-DLC Discovery in an AWS Quick team Space instead of a local folder, so several people can follow, open, and take over the same Discovery from their own AWS Quick Desktop. AWS Quick only, and optional: without a Space, Discovery runs from a local folder as usual. Use when the team wants to share Discovery in a Space, take over from another driver, or run pods in one Space.
Use case scoring for AI-DLC Discovery: ranks 3 to 5 candidate use cases, written as problems, with a gate, six weighted dimensions, a confidence rating, and a decision (build first, prototype to learn, defer, drop), then helps the team select what to take forward. Also re-scores after Validate. Use when the user wants to prioritize or score use cases, decide which problem to work on first, or revisit scores.
Validate phase of AI-DLC Discovery: challenges the selected use case on the day against four risks (valuable, usable, feasible, viable), runs a pre-mortem, sets testable success criteria, prioritizes the customer problems with the hypotheses still open, and records the team's decision: commit, pivot, or park. Use when the user wants to validate a concept, challenge assumptions, run a pre-mortem, set success criteria, or decide whether to build.
Working Backwards for AI-DLC Discovery: runs the five Working Backwards questions with the team, then writes the PR/FAQ, a readiness check, the PRD, and the riskiest assumption for a selected use case. Use when the user wants to write or challenge a PR/FAQ or press release, write a PRD, run Working Backwards, or name what the prototype must prove.